AI-generated patient letters

The comprehensibility of hospital discharge letters has a direct impact on patients' subsequent behaviour during aftercare. Unclear medical wording or a lack of advice on lifestyle make it difficult for them to actively participate in their treatment and implement the recommended measures. This can lead to misunderstandings, lower adherence to treatment and potentially avoidable post-treatment costs.

AI systems offer the opportunity to present complex information in a more comprehensible way and thus increase patient activation. Patient activation describes the degree to which patients have the knowledge, skills and self-confidence to actively participate in their healthcare.

1. Improved comprehensibility
   - Use of large language models to automatically translate medical terms and complex issues into patient-friendly language.
2. Addition of lifestyle recommendations
   - Automatic addition of personalised lifestyle recommendations on nutrition, exercise and prevention, etc., so that patients receive specific recommendations for action directly in the medical report.
3. Impact measurement
   - Investigation into whether the AI-generated patient letters are technically correct, complete and secure.
   - Analysis of the extent to which these improved letters can increase patient activation.

• Funding: The project is funded by the internal research funding of Witten/Herdecke University. Funding reference number: IFF 2024-80.
• Responsible: Chair of Health Informatics & Chair of Internal Medicine I